2025 Google Bananza - A year in review under the sign of the banana

When you hear "2025 Google Bananza", your thoughts quickly turn to the Ponderosa Ranch: Ben Cartwright, his sons, the Wild West. The name is a play on words from Banana and Bonanza - and pretty much sums up what it was all about: a review of the year under the sign of a yellow, crooked superstar.

But instead of revolvers and riding boots, aprons, wooden spoons and a few charts dominated the scene. At the Google Sihlpost, Pia De Carli and Samuel Leiser from the Google Switzerland communications team gave a relaxed and humorous tour of the late morning - more like the hosts of a lunch among friends than the moderating duo of a tech show. At the center of the short prelude: the new AI image model Nano Banana Proembedded in a setting that smelled like a cooking show, not a developer conference.

Bananza on the 4th floor: Between banana cake and beta features

The show kitchen on the 4th floor of Kasernenstrasse felt less like a "press event" and more like a studio for a food format. While generative AI, Gemini 3 and new workflows were briefly discussed beforehand, things later sizzled and bubbled with chef Harris, who transformed the morning into a multi-sensory experience.

On the menu: a creamy pumpkin risotto with fine meat - and, nomen est omen, a banana cake as a reference to the AI model. No PowerPoint battles, no laser pointers, just pots, pans and fresh ingredients.

This setting is the central theme of the three-and-a-half-minute video about the event: Close-ups of cutting, stirring and tasting, plus input from the Google team on Nano Banana Pro. The contrast could hardly have been clearer: Here craftsmanship and heat, there computing power and pixel precision.

The secret star: Nano Banana Pro

The banana was more than just decoration this morning. The real protagonist was called Nano Banana Pro - Google's new AI image model, which is based on Gemini 3 Pro and was rolled out on November 20. Target group: Marketing professionals, agencies, creators - anyone who doesn't just need "any" image, but exactly the visual that fits the concept.

Marta Zaratin, AI Marketing Manager EMEA, formulated the starting point as follows: "The picture is there, but it's not there yet. perfect." This is exactly where Nano Banana Pro comes in. The step forward is not just about creating images. Generatebut to process them in dialog - layer by layer, level by level, almost like in a conversation with an art-directing AI partner.

Text that finally sits

Anyone who has ever tried to use an image AI to produce a poster with a claim will know the drama: a jumble of letters, typography that looks out of place, logos that look like they come from a parallel universe.

Nano Banana Pro should close this chapter. The model renders texts cleanly, legibly and consistently - and in multiple languages. Whether it's a slogan on a drinks can that is correctly translated into Korean for the Asian market or "BERLIN" lettering that is subtly formed from building facades: The AI understands language, context and design at the same time.

For creatives, this means less frustration in the finishing touches, fewer manual corrections - and more time for what cannot be automated: the idea behind it.

Consistency instead of a grab bag

Another pain point in the industry: storytelling with AI. When protagonists look different in every scene, the campaign becomes a grab bag. This is exactly where Nano Banana Pro wants to deliver.

The model can link up to 14 images and keep the similarity of up to five people or characters consistent. Google showed a group of 14 fluffy monsters on a sofa. Each had its own character - and yet remained recognizable across different shots.

For brands with recurring characters, social series or campaigns with recurring scenes, this is a massive step forward. A "one-off AI stroke of luck" becomes a plannable visual world.

Light, depth of field, mood: fine-tuning by prompt

In addition, there is a level of detail control that until recently was reserved for Photoshop professionals. A prompt such as "Turn this scene into nighttime" is enough to transform a sunny scene into a night shot: streetlights come on, reflections change, shadows become longer.

Controlling the depth of field works in a similar way: with a "focus on the flowers", the audience's gaze can be directed specifically to one element, the background becomes softer and the product moves to the front.

For campaign production, this means: layout variants in seconds, faster visual AB tests, easier translation of a visual idea into different channels and formats.

From pitch decks to CVs: AI is becoming an everyday tool

In addition to these "hero features", Google presented a list of fields of application that clearly shows how deeply AI is already seeping into everyday working life. Nano Banana Pro and the associated tools help to visualize presentations, build flyers and social assets, sketch logos, enhance application photos or even illustrate trading scenarios.

The function is exciting "Real-world knowledge"AI recognizes ingredients in a picture and turns them into a recipe - including an attractive visual. Or it takes current weather data and transforms it into a pop-art infographic that is displayed directly in the social feed.

What used to require several tools, stock portals and a design round can now be initiated in dialog with the model. The real question for agencies and marketing departments is shifting from "What is technically possible?" to "How do we organize our work around these possibilities?".

SynthID: Transparency in the pixel

An invisible but crucial layer runs in the background: SynthIDGoogle's digital watermark. All images generated with the AI are marked with it without the human eye seeing anything.

For an industry that works with trust, responsibility and brand values, this is more than just a technical detail. It is a signal that AI visuals should not be dumped into the content stream as an anonymous mass, but as clearly declared building blocks of a communication strategy.

Conclusion at the stove: The machine paints, Harris cooks

After lunch, a simple realization was on the table - alongside the pumpkin risotto and the banana cake: Nano Banana Pro can create the most photorealistic image of a dessert in seconds, including perfectly placed lettering, ideal icing and social-ready lighting mood. The recipe can be prepared as an infographic that rivals any food blog.

But what AI can't do: create the smell in the kitchen, the clinking of glasses, the brief moment when everyone takes their first bite at the same time. It was precisely these analog moments that made the "2025 Google Bananza" more than just a product demo.

For the marketing, media and creative industries, the message is: the tools are spectacularly strong. Texts fit, visual worlds remain consistent, light and depth of field follow the prompt. But the decision as to which stories are told - and how they should feel in real life - remains with people.

The fact that Google preferred to use a wooden spoon in Zurich instead of just serving slides fits in perfectly. The actual bananza has long been taking place in the data centers. But the place where communication really works is still where someone stands at the stove, creates something - and lets others participate.